Disease Prediction¶

Importing libraries¶

In [5]:
import pandas as pd
import matplotlib.pyplot as plt
import seaborn as sns
import numpy as np
from sklearn.linear_model import LogisticRegression
from sklearn.tree import DecisionTreeClassifier
from sklearn.ensemble import RandomForestClassifier
from sklearn.tree import export_graphviz #plot tree
from sklearn.metrics import roc_curve, auc #for model evaluation
from sklearn.metrics import classification_report #for model evaluation
from sklearn.metrics import confusion_matrix #for model evaluation
from sklearn.model_selection import train_test_split #for data splitting
from sklearn.metrics import r2_score,accuracy_score
from sklearn.svm import SVC
from sklearn.naive_bayes import GaussianNB
from sklearn.preprocessing import LabelEncoder
from sklearn.metrics import accuracy_score

from sklearn.tree import export_graphviz #plot tree
from sklearn.metrics import roc_curve, auc #for model evaluation
from sklearn.metrics import classification_report #for model evaluation
from sklearn.metrics import confusion_matrix #for model evaluation

import pylab
import warnings
warnings.filterwarnings('ignore')
%matplotlib inline

Reading Train Dataset¶

In [6]:
train_set = pd.read_csv('Training.csv')
test_set = pd.read_csv('Testing.csv')
train_set = train_set.iloc[:,:-1]
train_set.head()
Out[6]:
itching skin_rash nodal_skin_eruptions continuous_sneezing shivering chills joint_pain stomach_pain acidity ulcers_on_tongue ... blackheads scurring skin_peeling silver_like_dusting small_dents_in_nails inflammatory_nails blister red_sore_around_nose yellow_crust_ooze prognosis
0 1 1 1 0 0 0 0 0 0 0 ... 0 0 0 0 0 0 0 0 0 Fungal infection
1 0 1 1 0 0 0 0 0 0 0 ... 0 0 0 0 0 0 0 0 0 Fungal infection
2 1 0 1 0 0 0 0 0 0 0 ... 0 0 0 0 0 0 0 0 0 Fungal infection
3 1 1 0 0 0 0 0 0 0 0 ... 0 0 0 0 0 0 0 0 0 Fungal infection
4 1 1 1 0 0 0 0 0 0 0 ... 0 0 0 0 0 0 0 0 0 Fungal infection

5 rows × 133 columns

Shape of training set¶

In [7]:
nRow,nCol=train_set.shape
print(f'There are {nRow} rows and {nCol} columns. Prognosis is the target/label variable.')

There are 4920 rows and 133 columns. Prognosis is the target/label variable.

Output Variable: Prognosis¶

In [8]:
train_set['prognosis'].value_counts()
Out[8]:
Fungal infection                           120
Hepatitis C                                120
Hepatitis E                                120
Alcoholic hepatitis                        120
Tuberculosis                               120
Common Cold                                120
Pneumonia                                  120
Dimorphic hemmorhoids(piles)               120
Heart attack                               120
Varicose veins                             120
Hypothyroidism                             120
Hyperthyroidism                            120
Hypoglycemia                               120
Osteoarthristis                            120
Arthritis                                  120
(vertigo) Paroymsal  Positional Vertigo    120
Acne                                       120
Urinary tract infection                    120
Psoriasis                                  120
Hepatitis D                                120
Hepatitis B                                120
Allergy                                    120
hepatitis A                                120
GERD                                       120
Chronic cholestasis                        120
Drug Reaction                              120
Peptic ulcer diseae                        120
AIDS                                       120
Diabetes                                   120
Gastroenteritis                            120
Bronchial Asthma                           120
Hypertension                               120
Migraine                                   120
Cervical spondylosis                       120
Paralysis (brain hemorrhage)               120
Jaundice                                   120
Malaria                                    120
Chicken pox                                120
Dengue                                     120
Typhoid                                    120
Impetigo                                   120
Name: prognosis, dtype: int64
In [9]:
print(f'**Summary**:\n There are 41 diseases in the dataset and each containing 120 rows. So, the dataset is equally balanced.')

**Summary**:
 There are 41 diseases in the dataset and each containing 120 rows. So, the dataset is equally balanced.

Data Pre-processing¶

Checking Missing data points¶

In [10]:
train_set.isna().sum()
Out[10]:
itching                 0
skin_rash               0
nodal_skin_eruptions    0
continuous_sneezing     0
shivering               0
                       ..
inflammatory_nails      0
blister                 0
red_sore_around_nose    0
yellow_crust_ooze       0
prognosis               0
Length: 133, dtype: int64
In [11]:
test_set.isna().sum()
Out[11]:
itching                 0
skin_rash               0
nodal_skin_eruptions    0
continuous_sneezing     0
shivering               0
                       ..
inflammatory_nails      0
blister                 0
red_sore_around_nose    0
yellow_crust_ooze       0
prognosis               0
Length: 133, dtype: int64
In [12]:
symtom_dict = {}
for index,column in enumerate(train_set.columns):
    symtom_dict[column] = index
    


train_set['prognosis'].replace(
    {}, inplace=True)
In [13]:
train_set['prognosis']
Out[13]:
0                              Fungal infection
1                              Fungal infection
2                              Fungal infection
3                              Fungal infection
4                              Fungal infection
                         ...                   
4915    (vertigo) Paroymsal  Positional Vertigo
4916                                       Acne
4917                    Urinary tract infection
4918                                  Psoriasis
4919                                   Impetigo
Name: prognosis, Length: 4920, dtype: object
In [14]:
for col in train_set.columns:
    if col =='prognosis':
        continue
    sns.countplot(data = train_set , x = col)
    plt.show()
In [15]:
# Get the number of missing data points, NA's ,NAN's values per column
total = train_set.isnull().sum().sort_values(ascending=False)
percent = (train_set.isnull().sum()/train_set.isnull().count()).sort_values(ascending=False)
missing_data = pd.concat([total, percent], axis=1, keys=['Total', 'Percent'])

total = train_set.isna().sum().sort_values(ascending=False)
percent = (train_set.isna().sum()/train_set.isna().count()).sort_values(ascending=False)
na_data = pd.concat([total, percent], axis=1, keys=['Total', 'Percent'])

if((na_data.all()).all()>0 or (na_data.all()).all()>0):
     print('Found Missing Data or NA values')
        
else:
    print('There is no missing data or null values in the collected data. Additionally, the length of each column is same.')

There is no missing data or null values in the collected data. Additionally, the length of each column is same.

Detecting Outliers¶

In [16]:
temp_df=train_set.iloc[:,:-1]
#Detect outliers
plt.subplots(figsize=(18,10))
temp_df.iloc[:,:50].boxplot()
plt.xticks(rotation=90)
plt.show()

plt.subplots(figsize=(18,10))
temp_df.iloc[:,50:].boxplot()
plt.xticks(rotation=90)
plt.show()

print(f'**Summary**:\n No outliers')

**Summary**:
 No outliers

Correlation Matrix¶

In [17]:
plt.figure(figsize = (30, 30))
sns.heatmap(train_set.corr(), cmap = 'PuBu', annot = False)
plt.show()

Dropping feature columns with correlation greater than 0.9¶

In [18]:
corr_matrix=train_set.corr()
upper = corr_matrix.where(np.triu(np.ones(corr_matrix.shape), k=1).astype(np.bool))
upper
Out[18]:
itching skin_rash nodal_skin_eruptions continuous_sneezing shivering chills joint_pain stomach_pain acidity ulcers_on_tongue ... pus_filled_pimples blackheads scurring skin_peeling silver_like_dusting small_dents_in_nails inflammatory_nails blister red_sore_around_nose yellow_crust_ooze
itching NaN 0.318158 0.326439 -0.086906 -0.059893 -0.175905 -0.160650 0.202850 -0.086906 -0.059893 ... -0.059893 -0.059893 -0.059893 -0.061573 -0.061573 -0.061573 -0.061573 -0.061573 -0.061573 -0.061573
skin_rash NaN NaN 0.298143 -0.094786 -0.065324 -0.029324 0.171134 0.161784 -0.094786 -0.065324 ... 0.320859 0.320859 0.320859 0.331087 0.331087 0.331087 0.331087 0.331087 0.331087 0.331087
nodal_skin_eruptions NaN NaN NaN -0.032566 -0.022444 -0.065917 -0.060200 -0.032566 -0.032566 -0.022444 ... -0.022444 -0.022444 -0.022444 -0.023073 -0.023073 -0.023073 -0.023073 -0.023073 -0.023073 -0.023073
continuous_sneezing NaN NaN NaN NaN 0.608981 0.446238 -0.087351 -0.047254 -0.047254 -0.032566 ... -0.032566 -0.032566 -0.032566 -0.033480 -0.033480 -0.033480 -0.033480 -0.033480 -0.033480 -0.033480
shivering NaN NaN NaN NaN NaN 0.295332 -0.060200 -0.032566 -0.032566 -0.022444 ... -0.022444 -0.022444 -0.022444 -0.023073 -0.023073 -0.023073 -0.023073 -0.023073 -0.023073 -0.023073
... ... ... ... ... ... ... ... ... ... ... ... ... ... ... ... ... ... ... ... ... ...
small_dents_in_nails NaN NaN NaN NaN NaN NaN NaN NaN NaN NaN ... NaN NaN NaN NaN NaN NaN 0.946120 -0.023720 -0.023720 -0.023720
inflammatory_nails NaN NaN NaN NaN NaN NaN NaN NaN NaN NaN ... NaN NaN NaN NaN NaN NaN NaN -0.023720 -0.023720 -0.023720
blister NaN NaN NaN NaN NaN NaN NaN NaN NaN NaN ... NaN NaN NaN NaN NaN NaN NaN NaN 0.946120 0.946120
red_sore_around_nose NaN NaN NaN NaN NaN NaN NaN NaN NaN NaN ... NaN NaN NaN NaN NaN NaN NaN NaN NaN 0.946120
yellow_crust_ooze NaN NaN NaN NaN NaN NaN NaN NaN NaN NaN ... NaN NaN NaN NaN NaN NaN NaN NaN NaN NaN

132 rows × 132 columns

In [19]:
# Find index of feature columns with correlation greater than 0.9
to_drop = [column for column in upper.columns if any(upper[column] > 0.9)]
print(to_drop,len(to_drop))

train_set=train_set.drop(to_drop, axis=1)
test_set=test_set.drop(to_drop, axis=1)

['cold_hands_and_feets', 'redness_of_eyes', 'sinus_pressure', 'runny_nose', 'congestion', 'pain_in_anal_region', 'bloody_stool', 'irritation_in_anus', 'bruising', 'swollen_legs', 'swollen_blood_vessels', 'puffy_face_and_eyes', 'enlarged_thyroid', 'brittle_nails', 'swollen_extremeties', 'drying_and_tingling_lips', 'slurred_speech', 'hip_joint_pain', 'unsteadiness', 'loss_of_smell', 'continuous_feel_of_urine', 'internal_itching', 'altered_sensorium', 'belly_pain', 'abnormal_menstruation', 'increased_appetite', 'polyuria', 'receiving_blood_transfusion', 'receiving_unsterile_injections', 'coma', 'stomach_bleeding', 'distention_of_abdomen', 'history_of_alcohol_consumption', 'fluid_overload.1', 'prominent_veins_on_calf', 'palpitations', 'painful_walking', 'silver_like_dusting', 'small_dents_in_nails', 'inflammatory_nails', 'red_sore_around_nose', 'yellow_crust_ooze'] 42

Dropping constant feature columns¶

Label Encoder¶

In [21]:
encoder = LabelEncoder()
train_set["prognosis"] = encoder.fit_transform(train_set["prognosis"])
test_set["prognosis"] = encoder.transform(test_set["prognosis"])

Test Valid Split¶

In [22]:
X_train, X_valid, y_train, y_valid = train_test_split(train_set.drop('prognosis', 1), train_set['prognosis'], test_size = .4, random_state=42,shuffle=True)
In [23]:
X_train.shape
Out[23]:
(2952, 90)
In [24]:
test_set = pd.concat([test_set,pd.concat([X_valid,y_valid],axis=1)],axis=0)
test_set.shape
Out[24]:
(2010, 91)

Logistic Regression¶

In [25]:
lr=LogisticRegression(C=0.2,random_state=42, penalty='l2')
lr.fit(X_train,y_train)
print("Logistic Train score with ",format(lr.score(X_train, y_train)))

Logistic Train score with  1.0
In [26]:
print("Logistic Test score with ",format(lr.score(test_set.iloc[:,:-1], test_set['prognosis'])))

Logistic Test score with  0.9995024875621891

Confusion Matrix¶

In [27]:
y_pred = lr.predict(test_set.iloc[:,:-1])
class_names=encoder.classes_
fig, ax = plt.subplots(figsize = (20,10))
tick_marks = np.arange(len(class_names))
plt.xticks(tick_marks, class_names)
plt.yticks(tick_marks, class_names)
cm = confusion_matrix(test_set['prognosis'], y_pred)
sns.heatmap(cm, annot=True, cmap="YlGnBu" ,fmt='g')
ax.xaxis.set_label_position("top")
plt.tight_layout()
plt.title('Confusion matrix', y=1.1)
plt.ylabel('Actual label')
plt.xlabel('Predicted label')
Out[27]:
Text(0.5, 885.5555555555555, 'Predicted label')

Classification report¶

In [28]:
print(classification_report( test_set['prognosis'], y_pred))

              precision    recall  f1-score   support

           0       1.00      1.00      1.00        42
           1       1.00      1.00      1.00        47
           2       1.00      1.00      1.00        53
           3       1.00      1.00      1.00        56
           4       1.00      1.00      1.00        42
           5       1.00      1.00      1.00        47
           6       1.00      1.00      1.00        53
           7       1.00      1.00      1.00        46
           8       1.00      1.00      1.00        46
           9       1.00      1.00      1.00        48
          10       1.00      1.00      1.00        47
          11       1.00      1.00      1.00        59
          12       1.00      1.00      1.00        42
          13       1.00      1.00      1.00        55
          14       1.00      1.00      1.00        49
          15       1.00      0.98      0.99        51
          16       1.00      1.00      1.00        51
          17       1.00      1.00      1.00        56
          18       1.00      1.00      1.00        51
          19       1.00      1.00      1.00        55
          20       1.00      1.00      1.00        46
          21       1.00      1.00      1.00        47
          22       1.00      1.00      1.00        58
          23       1.00      1.00      1.00        50
          24       1.00      1.00      1.00        50
          25       1.00      1.00      1.00        52
          26       1.00      1.00      1.00        50
          27       1.00      1.00      1.00        55
          28       1.00      1.00      1.00        46
          29       1.00      1.00      1.00        39
          30       1.00      1.00      1.00        43
          31       1.00      1.00      1.00        38
          32       1.00      1.00      1.00        49
          33       1.00      1.00      1.00        38
          34       1.00      1.00      1.00        54
          35       0.98      1.00      0.99        48
          36       1.00      1.00      1.00        53
          37       1.00      1.00      1.00        50
          38       1.00      1.00      1.00        50
          39       1.00      1.00      1.00        48
          40       1.00      1.00      1.00        50

    accuracy                           1.00      2010
   macro avg       1.00      1.00      1.00      2010
weighted avg       1.00      1.00      1.00      2010

Decision Tree¶

In [29]:
dt = DecisionTreeClassifier()
dt.fit(X_train, y_train)
print("Decision Tree Train score with ",format(dt.score(X_train, y_train)))

Decision Tree Train score with  1.0
In [30]:
print("Decision Tree Test score with ",format(dt.score(test_set.iloc[:,:-1], test_set['prognosis'])))

Decision Tree Test score with  0.9965174129353234

Confusion Matrix¶

In [31]:
y_pred = dt.predict(test_set.iloc[:,:-1])
class_names=encoder.classes_
fig, ax = plt.subplots(figsize = (20,10))
tick_marks = np.arange(len(class_names))
plt.xticks(tick_marks, class_names)
plt.yticks(tick_marks, class_names)
cm = confusion_matrix(test_set['prognosis'], y_pred)
sns.heatmap(cm, annot=True, cmap="YlGnBu" ,fmt='g')
ax.xaxis.set_label_position("top")
plt.tight_layout()
plt.title('Confusion matrix', y=1.1)
plt.ylabel('Actual label')
plt.xlabel('Predicted label')
Out[31]:
Text(0.5, 885.5555555555555, 'Predicted label')

Classification report¶

In [32]:
print(classification_report( test_set['prognosis'], y_pred))

              precision    recall  f1-score   support

           0       1.00      1.00      1.00        42
           1       1.00      1.00      1.00        47
           2       1.00      1.00      1.00        53
           3       1.00      1.00      1.00        56
           4       1.00      1.00      1.00        42
           5       1.00      1.00      1.00        47
           6       0.98      1.00      0.99        53
           7       1.00      1.00      1.00        46
           8       1.00      1.00      1.00        46
           9       1.00      1.00      1.00        48
          10       1.00      1.00      1.00        47
          11       1.00      1.00      1.00        59
          12       1.00      1.00      1.00        42
          13       1.00      1.00      1.00        55
          14       1.00      1.00      1.00        49
          15       1.00      0.98      0.99        51
          16       1.00      1.00      1.00        51
          17       1.00      1.00      1.00        56
          18       1.00      1.00      1.00        51
          19       1.00      1.00      1.00        55
          20       1.00      1.00      1.00        46
          21       1.00      1.00      1.00        47
          22       1.00      1.00      1.00        58
          23       1.00      1.00      1.00        50
          24       1.00      0.88      0.94        50
          25       1.00      1.00      1.00        52
          26       1.00      1.00      1.00        50
          27       1.00      1.00      1.00        55
          28       1.00      1.00      1.00        46
          29       1.00      1.00      1.00        39
          30       1.00      1.00      1.00        43
          31       1.00      1.00      1.00        38
          32       1.00      1.00      1.00        49
          33       1.00      1.00      1.00        38
          34       0.90      1.00      0.95        54
          35       1.00      1.00      1.00        48
          36       1.00      1.00      1.00        53
          37       1.00      1.00      1.00        50
          38       1.00      1.00      1.00        50
          39       1.00      1.00      1.00        48
          40       1.00      1.00      1.00        50

    accuracy                           1.00      2010
   macro avg       1.00      1.00      1.00      2010
weighted avg       1.00      1.00      1.00      2010

Random Forest¶

In [33]:
rf = RandomForestClassifier(max_depth=6,oob_score=True,random_state=42,criterion='entropy',max_features='auto',n_estimators=300)
rf.fit(X_train, y_train)
y_pred=rf.predict(X_valid)
print("Random Forest Train score with ",format(rf.score(X_train, y_train)))

Random Forest Train score with  0.9969512195121951
In [34]:
print("Random Forest Test score with ",format(rf.score(test_set.iloc[:,:-1], test_set['prognosis'])))

Random Forest Test score with  0.9955223880597015

Confusion Matrix¶

In [35]:
y_pred = rf.predict(test_set.iloc[:,:-1])
class_names=encoder.classes_
fig, ax = plt.subplots(figsize = (20,10))
tick_marks = np.arange(len(class_names))
plt.xticks(tick_marks, class_names)
plt.yticks(tick_marks, class_names)
cm = confusion_matrix(test_set['prognosis'], y_pred)
sns.heatmap(cm, annot=True, cmap="YlGnBu" ,fmt='g')
ax.xaxis.set_label_position("top")
plt.tight_layout()
plt.title('Confusion matrix', y=1.1)
plt.ylabel('Actual label')
plt.xlabel('Predicted label')
Out[35]:
Text(0.5, 885.5555555555555, 'Predicted label')

Classification report¶

In [36]:
print(classification_report( test_set['prognosis'], y_pred))

              precision    recall  f1-score   support

           0       1.00      1.00      1.00        42
           1       0.90      1.00      0.95        47
           2       1.00      0.96      0.98        53
           3       1.00      1.00      1.00        56
           4       1.00      1.00      1.00        42
           5       1.00      1.00      1.00        47
           6       1.00      1.00      1.00        53
           7       1.00      1.00      1.00        46
           8       1.00      1.00      1.00        46
           9       1.00      1.00      1.00        48
          10       1.00      1.00      1.00        47
          11       1.00      1.00      1.00        59
          12       1.00      1.00      1.00        42
          13       1.00      1.00      1.00        55
          14       1.00      1.00      1.00        49
          15       1.00      1.00      1.00        51
          16       1.00      1.00      1.00        51
          17       1.00      1.00      1.00        56
          18       1.00      1.00      1.00        51
          19       1.00      1.00      1.00        55
          20       1.00      1.00      1.00        46
          21       0.96      1.00      0.98        47
          22       1.00      0.97      0.98        58
          23       1.00      1.00      1.00        50
          24       1.00      1.00      1.00        50
          25       1.00      1.00      1.00        52
          26       1.00      1.00      1.00        50
          27       1.00      0.91      0.95        55
          28       1.00      1.00      1.00        46
          29       1.00      1.00      1.00        39
          30       1.00      1.00      1.00        43
          31       1.00      1.00      1.00        38
          32       1.00      1.00      1.00        49
          33       1.00      1.00      1.00        38
          34       1.00      1.00      1.00        54
          35       1.00      1.00      1.00        48
          36       1.00      1.00      1.00        53
          37       1.00      1.00      1.00        50
          38       0.96      1.00      0.98        50
          39       1.00      1.00      1.00        48
          40       1.00      1.00      1.00        50

    accuracy                           1.00      2010
   macro avg       1.00      1.00      1.00      2010
weighted avg       1.00      1.00      1.00      2010

SVM¶

In [37]:
svm = SVC()
svm.fit(X_train, y_train)
y_pred=svm.predict(X_valid)
print("SVM Train score with ",format(svm.score(X_train, y_train)))

SVM Train score with  1.0
In [38]:
print("SVM Test score with ",format(svm.score(test_set.iloc[:,:-1], test_set['prognosis'])))

SVM Test score with  0.9995024875621891

Confusion Matrix¶

In [39]:
y_pred = svm.predict(test_set.iloc[:,:-1])
class_names=encoder.classes_
fig, ax = plt.subplots(figsize = (20,10))
tick_marks = np.arange(len(class_names))
plt.xticks(tick_marks, class_names)
plt.yticks(tick_marks, class_names)
cm = confusion_matrix(test_set['prognosis'], y_pred)
sns.heatmap(cm, annot=True, cmap="YlGnBu" ,fmt='g')
ax.xaxis.set_label_position("top")
plt.tight_layout()
plt.title('Confusion matrix', y=1.1)
plt.ylabel('Actual label')
plt.xlabel('Predicted label')
Out[39]:
Text(0.5, 885.5555555555555, 'Predicted label')

Classification report¶

In [40]:
print(classification_report( test_set['prognosis'], y_pred))

              precision    recall  f1-score   support

           0       1.00      1.00      1.00        42
           1       1.00      1.00      1.00        47
           2       1.00      1.00      1.00        53
           3       1.00      1.00      1.00        56
           4       1.00      1.00      1.00        42
           5       1.00      1.00      1.00        47
           6       1.00      1.00      1.00        53
           7       1.00      1.00      1.00        46
           8       1.00      1.00      1.00        46
           9       1.00      1.00      1.00        48
          10       1.00      1.00      1.00        47
          11       1.00      1.00      1.00        59
          12       1.00      1.00      1.00        42
          13       1.00      1.00      1.00        55
          14       1.00      1.00      1.00        49
          15       1.00      0.98      0.99        51
          16       1.00      1.00      1.00        51
          17       1.00      1.00      1.00        56
          18       1.00      1.00      1.00        51
          19       1.00      1.00      1.00        55
          20       1.00      1.00      1.00        46
          21       1.00      1.00      1.00        47
          22       1.00      1.00      1.00        58
          23       1.00      1.00      1.00        50
          24       1.00      1.00      1.00        50
          25       1.00      1.00      1.00        52
          26       1.00      1.00      1.00        50
          27       1.00      1.00      1.00        55
          28       1.00      1.00      1.00        46
          29       1.00      1.00      1.00        39
          30       1.00      1.00      1.00        43
          31       1.00      1.00      1.00        38
          32       1.00      1.00      1.00        49
          33       1.00      1.00      1.00        38
          34       1.00      1.00      1.00        54
          35       0.98      1.00      0.99        48
          36       1.00      1.00      1.00        53
          37       1.00      1.00      1.00        50
          38       1.00      1.00      1.00        50
          39       1.00      1.00      1.00        48
          40       1.00      1.00      1.00        50

    accuracy                           1.00      2010
   macro avg       1.00      1.00      1.00      2010
weighted avg       1.00      1.00      1.00      2010

Naive Bayes¶

In [41]:
bayes = GaussianNB()
bayes.fit(X_train, y_train)
y_pred=bayes.predict(X_valid)
print("Naive Bayes Train score with ",format(bayes.score(X_train, y_train)))

Naive Bayes Train score with  1.0
In [42]:
print("Naive Bayes Test score with ",format(bayes.score(test_set.iloc[:,:-1], test_set['prognosis'])),'%')

Naive Bayes Test score with  0.9995024875621891 %

Confusion Matrix¶

In [43]:
y_pred = bayes.predict(test_set.iloc[:,:-1])
class_names=encoder.classes_
fig, ax = plt.subplots(figsize = (20,10))
tick_marks = np.arange(len(class_names))
plt.xticks(tick_marks, class_names)
plt.yticks(tick_marks, class_names)
cm = confusion_matrix(test_set['prognosis'], y_pred)
sns.heatmap(cm, annot=True, cmap="YlGnBu" ,fmt='g')
ax.xaxis.set_label_position("top")
plt.tight_layout()
plt.title('Confusion matrix', y=1.1)
plt.ylabel('Actual label')
plt.xlabel('Predicted label')
Out[43]:
Text(0.5, 885.5555555555555, 'Predicted label')

Classification report¶

In [44]:
print(classification_report( test_set['prognosis'], y_pred))

              precision    recall  f1-score   support

           0       1.00      1.00      1.00        42
           1       1.00      1.00      1.00        47
           2       1.00      1.00      1.00        53
           3       1.00      1.00      1.00        56
           4       1.00      1.00      1.00        42
           5       1.00      1.00      1.00        47
           6       1.00      1.00      1.00        53
           7       1.00      1.00      1.00        46
           8       1.00      1.00      1.00        46
           9       1.00      1.00      1.00        48
          10       1.00      1.00      1.00        47
          11       1.00      1.00      1.00        59
          12       1.00      1.00      1.00        42
          13       1.00      1.00      1.00        55
          14       1.00      1.00      1.00        49
          15       1.00      0.98      0.99        51
          16       1.00      1.00      1.00        51
          17       1.00      1.00      1.00        56
          18       1.00      1.00      1.00        51
          19       1.00      1.00      1.00        55
          20       1.00      1.00      1.00        46
          21       1.00      1.00      1.00        47
          22       1.00      1.00      1.00        58
          23       1.00      1.00      1.00        50
          24       1.00      1.00      1.00        50
          25       1.00      1.00      1.00        52
          26       1.00      1.00      1.00        50
          27       1.00      1.00      1.00        55
          28       1.00      1.00      1.00        46
          29       1.00      1.00      1.00        39
          30       1.00      1.00      1.00        43
          31       1.00      1.00      1.00        38
          32       1.00      1.00      1.00        49
          33       1.00      1.00      1.00        38
          34       1.00      1.00      1.00        54
          35       0.98      1.00      0.99        48
          36       1.00      1.00      1.00        53
          37       1.00      1.00      1.00        50
          38       1.00      1.00      1.00        50
          39       1.00      1.00      1.00        48
          40       1.00      1.00      1.00        50

    accuracy                           1.00      2010
   macro avg       1.00      1.00      1.00      2010
weighted avg       1.00      1.00      1.00      2010

Classification report¶

Model Accuracy Comparision¶

In [45]:
sns.set(style="darkgrid")

acc = pd.DataFrame({'Model':['Logistic','Decision Tree','Random Tree','SVM','Naive Byes'],'Accuracy':[98.3,98.2,97.9,98.5,96.7]})

# Set the figure size
fig, ax =  plt.subplots(figsize=(12, 7))
ax.set_ylim(96.5, 99)
# plot a bar chart
sns.barplot(
    x="Model", 
    y="Accuracy", 
    data=acc,  
    ci=None, 
    color='#69b3a2',
    orient = 'v');
plt.show()
In [46]:
import joblib as joblib
In [48]:
joblib.dump(y_pred,'trained_model')
Out[48]:
['trained_model']
In [49]:
predi= joblib.load('trained_model')
In [ ]: